PlugWiki

agent-platform-model-registry Skill (Google's own repo): Manage the Model Registry

Lets your assistant list, describe, upload, update and delete models in the Agent Platform Model Registry on Google Cloud.

Installs
8.7K
on skills.sh
Stars
21.8K
on GitHub
Updated
Nov 2026
last repo commit

This skill, from Google's own skills repo, drives the gcloud CLI and a small Python snippet to manage models in the Agent Platform Model Registry. It can list, describe, upload, update and delete models and search Model Garden publisher models.

Its safety tiers decide what needs your approval. Listing runs freely, upload and update need a Yes after a preview of the exact command, and delete needs typed confirmation. It bundles no scripts.

Good for

  • ML engineers who register and version models on Google Cloud
  • Platform teams cleaning up or auditing a model registry
  • Anyone who needs to find a Model Garden publisher model ID

Install agent-platform-model-registry

Claude Code

  1. 1

    Download the skill folder from https://github.com/google/skills/tree/HEAD/skills/cloud/agent-platform-model-registry.

  2. 2

    Put it at ~/.claude/skills/agent-platform-model-registry/ so that SKILL.md sits directly inside.

  3. 3

    Start a new session; the skill loads when its description matches the task.

Requires gcloud CLI, signed in, Python google-cloud-aiplatform library.

The skill runs gcloud and python3 commands against your Google Cloud project, so there is no Claude app upload route.

Built from the package data we hold. Not yet tested by us.

What's in the skill

When it loads

Agent Platform Model Registry Management. Use when you need to upload, list, describe, update, or delete machine learning models (and their versions) in the Agent Platform Model Registry. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform models.

Sections in SKILL.md

  1. Overview
  2. Safety & Confirmation Tiers (CRITICAL)
  3. Phase 0: Environment Setup & Parameter Resolution
  4. 1. Listing Models (Tier R)
  5. 2. Describing a Model (Tier R)
  6. 3. Uploading a Model (Tier M)
  7. Example: Uploading a Custom Model
  8. 4. Updating a Model (Tier M)
  9. 5. Deleting a Model (Tier D)
  10. 6. Searching Publisher Models (Tier R)

Key facts

Author
Google LLC (vendor's own repo, google/skills)
License
Apache-2.0 (LICENSE and README agree)
Installs on skills.sh
8,746 (Nov 2, 2026)
GitHub stars
21,764 for the whole google/skills repo (Nov 2, 2026)
Last repo commit
Nov 1, 2026
Audits on skills.sh
Gen Agent Trust Hub: pass, Socket: pass, Snyk: pass
What it includes
1 file including SKILL.md
Skill version
1.0.1 (SKILL.md metadata)
Needs
The gcloud CLI (gcloud ai models) and the Python google-cloud-aiplatform library for updates and name search

What it includes

  • Resolves project, region and model ID; always passes --region and --project, never global.

    It reuses values from earlier turns, allows one gcloud config lookup, then asks you for anything missing.

    Runs code · Phase 0: Parameters

  • gcloud ai models list with --limit=50, the 50 most recently updated models.

    A full inventory or a count pages through every model and can take a minute or more, so it only runs if you ask.

    Runs code · List models (Tier R)

  • gcloud ai models describe for a model or a specific version.

    Use MODEL_ID or MODEL_ID@VERSION_ID.

    Runs code · Describe a model (Tier R)

  • gcloud ai models upload with a display name, container image and optional artifact URI.

    Registers a new model or a new version when --parent-model is given. Needs your Yes to the exact command first.

    Runs code · Upload a model (Tier M)

  • Changes display name or description with the Python SDK; gcloud has no update command.

    Runs a python3 snippet using google.cloud.aiplatform. Needs your Yes to the exact script first.

    Runs code · Update a model (Tier M)

  • gcloud ai models delete removes a model and all its versions after typed confirmation.

    The SKILL.md calls it irreversible and says all versions must be undeployed from endpoints first.

    Delete · Delete a model (Tier D)

  • gcloud ai model-garden models list --model-filter to verify a Model Garden model ID.

    Used before generating model details, to get the exact publishers/... name.

    Runs code · Search publisher models (Tier R)

From the official docs

Permissions

Medium risk

It can upload, change and permanently delete models in your Google Cloud project. Read the exact command in each prompt before you approve.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • It can delete a model and all its versions

    gcloud ai models delete is irreversible per the SKILL.md. It requires a typed confirmation, so read the model ID in the prompt before you reply. github.com

  • Upload and update change your registry

    Both need a Yes after the assistant shows the exact command. Check the project, region and display name in it. github.com

  • Listing everything can be slow

    The SKILL.md caps lists at 50 and warns that a full list or a count can take over a minute in a large project. github.com

  • Cost

    The SKILL.md says registry operations provision no serving compute, so it reports $0.00 compute charges, and standard Cloud Storage pricing applies to model artifacts. github.com

Technical details
Turning on read-onlyNo switch. The SKILL.md splits actions into Tier R (list, describe, get), Tier M (upload, update) and Tier D (delete); only Tier R runs without asking.
Talks toGoogle Cloud Agent Platform APIs for your project and region
Scopes

  • Runs gcloud ai models list, describe, upload and delete in your Google Cloud project

  • Runs a python3 snippet that updates a model name or description

  • Searches Model Garden publisher models

Questions

What does agent-platform-model-registry do?

It lets your assistant manage models in the Agent Platform Model Registry on Google Cloud: list, describe, upload, update, delete, and search Model Garden publisher models. It is not for training, deploying to endpoints, or models outside the registry.

Is this Google's own skill?

Yes. It is in the google/skills repo, which holds Google's agent skills for Google products.

Will it delete or change models without asking?

The SKILL.md requires approval first. Upload and update need an interactive Yes/No that shows the exact command, and the agent must wait a turn before running it. Delete needs typed confirmation such as "I confirm" or "Yes, delete it", asked before any pre-flight checks. Listing and describing run without asking.

Does it cost money?

The SKILL.md says registry operations do not provision serving compute, so it reports $0.00 compute charges, and standard Cloud Storage pricing applies to model artifacts. It tells the agent not to call an estimate_cost tool for these actions.

How do I install it?

Use the install box above. The README installs skills from google/skills with npx skills add google/skills. The SKILL.md names the gcloud CLI and the google-cloud-aiplatform Python library but does not describe sign-in. Our advice: run gcloud auth login first and use a project that has Agent Platform access.

Does it work in the Claude app?

The install box has no Claude app upload tab, because the skill runs gcloud and Python commands on your machine. It fits an agent with a shell.

Is it safe?

It bundles no scripts, but it runs gcloud and Python commands against your Google Cloud project, and one of them deletes models for good. The SKILL.md's confirmation tiers limit that, but they are instructions to the assistant, so check each command before you say Yes.

Information

updated Nov 2, 2026

Repository
google/skills
Folder
skills/cloud/agent-platform-model-registry

google/skillsskills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3
  1. agent-platform-model-registry SKILL.md · github.com
  2. google/skills README · github.com
  3. google/skills marketplace.json · github.com

Updated Nov 2, 2026.